Glean, an enterprise AI platform, has become a major player in model routing — automatically selecting the most cost-effective AI model for each task — as frontier models grow expensive and open-weight alternatives gain traction. The company reached $300 million in annual recurring revenue this year and claims its routing system delivers 4x cost savings compared to using Claude alone, by directing simpler queries to cheaper models and reserving expensive frontier models for complex work. This shift reflects a broader enterprise trend away from reliance on single AI providers toward multi-model strategies that include open-source options, driven primarily by the need to control spiraling AI costs.
Amazon Bedrock's AgentCore Payments service is now generally available, allowing AI agents to autonomously pay for APIs, content, and services using integrated Coinbase and Stripe wallets. The service supports multiple payment protocols including x402 and Machine Payment Protocol, with configurable spending limits and comprehensive audit logging. Enterprises can now deploy agents that handle transactions independently while maintaining security controls and cost visibility across production workloads.
Amazon Quick embedded chat is a conversational AI interface that can be integrated into web applications, but it requires customization to match an organization's visual branding and communication style. The customization operates through two mechanisms: CSS styling and SDK frame options for visual theming (colors, layouts, removing default branding), and agent persona instructions plus content options for defining tone and response behavior. After configuration, the embedded chat appears as a native component of the application rather than an external tool, with responses tailored to organizational context and communication preferences.
Amazon Web Services describes AIDA, an AI-powered system for searching large contract repositories using retrieval-augmented generation with metadata filtering on Amazon Bedrock Knowledge Bases. The system uses implicit metadata pre-filtering before semantic search, followed by explicit application-layer constraints, to narrow results and reduce retrieval noise in legal documents. Enterprises can now query complex contracts in natural language with improved accuracy, though AWS emphasizes that AI-generated interpretations should still be reviewed by legal professionals.
Axonius, a SaaS asset intelligence platform, deployed AI agents on Amazon Bedrock AgentCore using a dedicated-runtime-per-tenant silo model to maintain tenant isolation, integrate with existing authentication systems, and track costs per customer. Each tenant runs a separate agent instance with isolated microVMs, IAM-based access control, and shared knowledge bases with metadata filtering. The architecture enables junior analysts to run complex risk analyses without senior analyst involvement, while maintaining security through infrastructure-level enforcement, cost tracking via CloudWatch token metrics, and integration with Axonius's existing VPC-isolated customer environments.
Hypercubic, a startup founded by former Apple engineers, raised $5.3 million to build AI agents that automatically map and rewrite legacy COBOL applications into modern programming languages. The company's agents can complete modernization projects in months rather than years by recovering buried business logic, generating documentation, and producing equivalent code in languages like Java. This addresses a critical industry problem as 90% of current COBOL developers are expected to retire within five to ten years, leaving financial institutions and large enterprises with aging, unmaintainable systems.
NLPatent rebranded as Clerq and launched agentic AI workflows that conduct patent research tasks automatically. The system completes full patentability analysis in approximately 10 minutes, compared to the days or weeks required through traditional methods involving junior associates or external search firms. Patent attorneys can now perform this work directly under their own oversight rather than delegating it, with integrations announced alongside RPX Corp. and Park IP to embed the research capability into their platforms.
EliseAI is in talks to raise $300 million at a $3.7 billion valuation with a16z and Bessemer, up from a $2.2 billion valuation in August 2025. The AI assistant company, which automates tenant communications and maintenance request routing for property managers, would gain $1.5 billion in value over 12 months if the round closes as reported. The funding reflects investor appetite for AI automation in unglamorous but sticky markets where incumbents like Yardi move slowly.
Linear analyzed AI adoption patterns across 127,000 users in their project management platform from January to June 2026, finding that AI-active users more than doubled in every function and coding agent teams tripled their weekly pull requests from 21 to 65. Teams using coding agents drove the 111% increase in overall pull requests since June 2024, though time spent on traditional tasks in Linear remained flat despite the new work layer AI created. The data suggests AI has accelerated output rather than saved time, with product managers and designers increasingly shipping code directly.
Slack has launched Atlas, an AI assistant that operates within Slack channels and direct messages to answer questions, automate tasks, and integrate with 200+ external tools. Atlas learns from team communication patterns without requiring setup or training, and works transparently in shared channels where multiple team members can see and build on its responses. The system operates on Slack's enterprise security infrastructure and can be customized by teams to create specialized AI coworkers for specific functions.
A new AI workspace platform has been developed to help researchers conduct reproducible scientific research. The platform integrates AI tools to streamline workflows and improve documentation standards. This enables scientists to more easily share, verify, and build upon each other's work.
Nous Research shipped Bot Mode for its Hermes Agent, turning each Hermes profile into a named bot that can chat, use memory/skills, and communicate via a persistent Agent Inbox. Hermes Agent v0.20.3 bundled Bot Mode and made it default-on inside Hermes Desktop, after a one-day public beta plugin. Users can now deploy on desktop at no license cost and organize multiple bots into group rosters with limited multi-round turn-taking, while enterprise buyers need their own controls because there is no admin console or centralized audit/policy layer.
NVIDIA uses ChatGPT Work to automate routine tasks, consolidate real-time information flows, and replicate effective processes across teams worldwide. The deployment reduces manual work and accelerates how teams handle signals across the organization. This enables NVIDIA to standardize and scale operational workflows more efficiently.
Every AI story that matters,
in your inbox by 8am.
TLDRocket reads all relevant sources, removes duplicate coverage, and summarises the
day in two minutes. Follow companies and topics for alerts, or get the
briefing in Slack. Free, no spam, unsubscribe anytime.
Reading TLDRocket needs no cookies, and the readership counts we rely on come from
our own cookieless analytics. Google Analytics is the exception: it sets cookies and
reports to Google, so it stays switched off until you allow it. You can change your
mind any time from “Cookie settings” in the footer.
Strictly necessary
Session security and form protection (tldrocket-session,
XSRF-TOKEN, 2 hours). The site cannot work without them,
so they need no consent.
Always on
Google Analytics 4 (_ga,
_ga_<id>, up to 2 years). Measures which
stories and sections readers use. Google acts as a third-party processor and may
store the data outside the EU. No advertising, no profiling, no data sold.
